Static Space · free · jerry3638

🧮 Expandable GPGPU / 可扩展 GPGPU

Vortex base + Gen0 RTL · target sketch ~100T FP32, ≤100W, 28nm, CUDA-compat path · FPGA later.
目标量级:约 100T FP32 / ≤100W / 28nm,CUDA 兼容路径,后续 FPGA。

Docs / 文档placeholder
RTL / Gen0 notesplaceholder
Author@jerry3638

Gradio Spaces on free CPU now need HF PRO. This Static page keeps the interactive planner without paid hardware.

FP32 Throughput Planner / 算力估算

Planning estimate only — not measured silicon. GEMM FLOPs ≈ 2·N³.

Click Estimate to compute.

Browser matmul micro-bench / 浏览器矩阵乘

Tiny Float32 matmul in JS (software stand-in on your device — not the chip).

Ready.

Target sketch: ~100T FP32 · ≤100W · 28nm · CUDA-compat · Vortex + Gen0 RTL